You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
If the .bak file are removed it works, but by default .bak file are used.
Step to reproduce:
Grab both influxdb 0.9.6.1 & 0.10.0-beta2:
$ wget https://s3.amazonaws.com/influxdb/influxdb_0.9.6.1_x86_64.tar.gz
$ wget https://influxdb.s3.amazonaws.com/influxdb-v0.10.0-beta2_linux_amd64.tar.gz
$ tar xvfz influxdb_0.9.6.1_x86_64.tar.gz
$ tar xvfz influxdb-v0.10.0-beta2_linux_amd64.tar.gz
Start 0.9.6.1 with default on empty database:
$ rm -fr ~/.influxdb/*
$ ./influxdb_0.9.6.1_x86_64/usr/bin/influxd
2016/01/28 09:42:15 InfluxDB starting, version 0.9.6.1, branch 0.9.6, commit 6d3a8603cfdaf1a141779ed88b093dcc5c528e5e, built 2015-12-10T23:40:23+0000
2016/01/28 09:42:15 Go version go1.4.2, GOMAXPROCS set to 4
2016/01/28 09:42:15 no configuration provided, using default settings
Using CLI, create a database and insert data:
$ ./influxdb_0.9.6.1_x86_64/usr/bin/influx
Connected to http://localhost:8086 version 0.9.6.1
InfluxDB shell 0.9.6.1
> create database test
> use test
Using database test
> insert metric value=42
> exit
$ ./usr/bin/influx_tsm -parallel ~/.influxdb/data/
b1 and bz1 shard conversion. [1/9025]
-----------------------------------
Data directory is: /root/.influxdb/data/
Databases specified: all
Database backups enabled: yes
Parallel mode enabled: yes
2 shard(s) detected, 2 non-TSM shards detected.
Database Retention Path Engine Size
_internal monitor /root/.influxdb/data/_internal/monitor/1 bz1 131072
test default /root/.influxdb/data/test/default/2 bz1 32768
These shards will be converted. Proceed? y/N: y
Conversion starting....
Backing up 2 databases...
Database _internal backed up (341.687µs)
Database test backed up (136.496µs)
Conversion of /root/.influxdb/data/_internal/monitor/1 successful (4.079804ms)
Conversion of /root/.influxdb/data/test/default/2 successful (521.546µs)
Summary statistics
========================================
Databases converted: 2
Shards converted: 2
TSM files created: 2
Points read: 694
Points written: 694
NaN filtered: 0
Inf filtered: 0
Points without fields filtered: 0
Disk usage pre-conversion (bytes): 163840
Disk usage post-conversion (bytes): 12506
Reduction factor: 99%
Bytes per TSM point: 18.02
Total conversion time: 5.209903ms
Start new verion of server
$ ./usr/bin/influxd
2016/01/28 09:43:44 InfluxDB starting, version v0.10.0-beta2, branch 0.10.0, commit 58e0eed9cb166e0b004e6b41a4d64c39d24b40f4, built 2016-01-23T02:41:00.429242
2016/01/28 09:43:44 Go version go1.4.3, GOMAXPROCS set to 4
2016/01/28 09:43:44 no configuration provided, using default settings
[meta] 2016/01/28 09:43:44 Starting meta service
[meta] 2016/01/28 09:43:44 Listening on HTTP: 127.0.0.1:8091
[metastore] 2016/01/28 09:43:44 Using data dir: /root/.influxdb/meta
[metastore] 2016/01/28 09:43:44 Node at localhost:8088 [Follower]
[metastore] 2016/01/28 09:43:46 Node at localhost:8088 [Leader]. peers=[localhost:8088]
[meta] 2016/01/28 09:43:46 127.0.0.1 - - [28/Jan/2016:09:43:46 +0000] GET /?index=0 HTTP/1.1 200 209 - Go 1.1 package http a08b1129-c5a3-11e5-8001-000000000000 992.248µs
[meta] 2016/01/28 09:43:46 127.0.0.1 - - [28/Jan/2016:09:43:46 +0000] POST /execute HTTP/1.1 200 28 - Go 1.1 package http a08b4f4e-c5a3-11e5-8003-000000000000 3.41668ms
[meta] 2016/01/28 09:43:46 127.0.0.1 - - [28/Jan/2016:09:43:46 +0000] GET /?index=18 HTTP/1.1 200 217 - Go 1.1 package http a08b4337-c5a3-11e5-8002-000000000000 4.939432ms
[store] 2016/01/28 09:43:46 Using data dir: /root/.influxdb/data
[tsm1wal] 2016/01/28 09:43:46 tsm1 WAL starting with 10485760 segment size
[tsm1wal] 2016/01/28 09:43:46 tsm1 WAL writing to /root/.influxdb/wal/_internal/monitor/1
[filestore]2016/01/28 09:43:46 /root/.influxdb/data/_internal/monitor/1/000000001-000000001.tsm (#0) opened in 73.712µs
[wal] 2016/01/28 09:43:46 WAL starting with 30720 ready series size, 0.50 compaction threshold, and 52428800 partition size threshold
[wal] 2016/01/28 09:43:46 WAL writing to /root/.influxdb/wal/_internal.bak/monitor/1
[tsm1wal] 2016/01/28 09:43:46 tsm1 WAL starting with 10485760 segment size
[tsm1wal] 2016/01/28 09:43:46 tsm1 WAL writing to /root/.influxdb/wal/test/default/2
[filestore]2016/01/28 09:43:46 /root/.influxdb/data/test/default/2/000000001-000000001.tsm (#0) opened in 44.93µs
[wal] 2016/01/28 09:43:46 WAL starting with 30720 ready series size, 0.50 compaction threshold, and 52428800 partition size threshold
[wal] 2016/01/28 09:43:46 WAL writing to /root/.influxdb/wal/test.bak/default/2
[handoff] 2016/01/28 09:43:46 Starting hinted handoff service
Check for data & insert one new value:
$ ./usr/bin/influx
Connected to http://localhost:8086 version v0.10.0-beta2
InfluxDB shell v0.10.0-beta2
> use test
Using database test
> select * from metric;
name: metric
------------
time value
1453974167488955239 42
> insert metric value=43
Backup is not possible:
$ ./usr/bin/influxd backup -database test /tmp/backup1
2016/01/28 09:45:10 backing up db=test since 0001-01-01 00:00:00 +0000 UTC
2016/01/28 09:45:10 backing up metastore to /tmp/backup1/meta.00
2016/01/28 09:45:10 backing up db=test rp=default shard=2 to /tmp/backup1/test.default.00002.00 since 0001-01-01 00:00:00 +0000 UTC
2016/01/28 09:45:10 backup complete
$ ls -l /tmp/backup1
total 4
-rw-r--r-- 1 root root 289 Jan 28 09:45 meta.00
-rw-r--r-- 1 root root 0 Jan 28 09:45 test.default.00002.00
On server console:
[snapshot] 2016/01/28 09:45:10 backup not implemented on bz1 engine
See that it still use .bak file (.bak file are openned, .bak are modified after convertion/new server start):
$ ./usr/bin/influx
Connected to http://localhost:8086 version v0.10.0-beta2
InfluxDB shell v0.10.0-beta2
> use test
Using database test
> select * from metric
name: metric
------------
time value
1453974167488955239 42
Backup now works:
$ ./usr/bin/influxd backup -database test /tmp/backup2
2016/01/28 09:48:33 backing up db=test since 0001-01-01 00:00:00 +0000 UTC
2016/01/28 09:48:33 backing up metastore to /tmp/backup2/meta.00
2016/01/28 09:48:33 backing up db=test rp=default shard=2 to /tmp/backup2/test.default.00002.00 since 0001-01-01 00:00:00 +0000 UTC
2016/01/28 09:48:33 backup complete
$ ls -l /tmp/backup2
total 8
-rw-r--r-- 1 root root 289 Jan 28 09:48 meta.00
-rw-r--r-- 1 root root 2048 Jan 28 09:48 test.default.00002.00
The text was updated successfully, but these errors were encountered:
@PierreF Thank you for reporting this with the amount of detail that you did. We've updated the conversion tool to backup the databases to a separate directory.
The problem was the backed-up directories were not being ignored because foo.bak could be a valid name for a database. The shard data then reported itself as belonging to the original database name. That's why your new points were getting written to the "backup" directory.
The tool will no longer allow databases to be backed up underneath the data directory, and the instructions/examples have been updated accordingly.
When following instruction from https://github.com/influxdata/influxdb/blob/master/cmd/influx_tsm/README.md to convert database from 0.9.6.1 to 0.10.0-beta2 / bz1 to tsm, the new server still use bz1 storage for .bak file.
If the .bak file are removed it works, but by default .bak file are used.
Step to reproduce:
Grab both influxdb 0.9.6.1 & 0.10.0-beta2:
Start 0.9.6.1 with default on empty database:
Using CLI, create a database and insert data:
As said in https://github.com/influxdata/influxdb/blob/master/cmd/influx_tsm/README.md :
Check for data & insert one new value:
Backup is not possible:
See that it still use .bak file (.bak file are openned, .bak are modified after convertion/new server start):
Backup now works:
The text was updated successfully, but these errors were encountered: